kagami/

normalize_version_words.rs

use crate::normalize_version::{bound_every_version, interval_to_requirement};

// Turns the English of hand-written ranges into operators. A `through` or `to` range becomes a
// hyphen range, which `normalize_version` handles afterwards.
//   `1.20.1 and earlier`                      ->  `<= 1.20.1`
//   `7.2.10 and below, 8.0.5 and below`       ->  `<= 7.2.10, <= 8.0.5`
//   `6.2 or newer`                            ->  `>= 6.2`
//   `2.0.0+`                                  ->  `>= 2.0.0`
//   `All versions prior to 1.7.14 and 1.8.7`  ->  `< 1.7.14, < 1.8.7`
//   `2.2.0 <= V <= 2.10.1`                    ->  `>= 2.2.0, <= 2.10.1`
//   `1.2.0 through 1.8.7`                     ->  `1.2.0 - 1.8.7`
//   `from 3.5.0 to 4.3.10`                    ->  `3.5.0 - 4.3.10`
//   `>= 1.0 and < 2.0`                        ->  `>= 1.0, < 2.0`
//   `>= 1.0.0, < 1.2.6`                       ->  `>= 1.0.0, < 1.2.6`  (nothing to do)
pub fn words_to_operators(range: &str) -> String {
    let range = strip_prefix_ignore_case(range, "all versions").unwrap_or(range);
    let range = strip_prefix_ignore_case(range, "from ").unwrap_or(range);

    let upper_bound_prefixes = [
        ("up to and including", "<="),
        ("up to", "<="),
        ("prior to", "<"),
        ("earlier than", "<"),
        ("before", "<"),
        ("until", "<"),
    ];
    for (phrase, operator) in upper_bound_prefixes {
        if let Some(versions) = strip_prefix_ignore_case(range, phrase) {
            return bound_every_version(versions, operator, "");
        }
    }

    let bound_suffixes = [
        ("and newer", ">="),
        ("or newer", ">="),
        ("and later", ">="),
        ("or later", ">="),
        ("and above", ">="),
        ("or above", ">="),
        ("and earlier", "<="),
        ("or earlier", "<="),
        ("and before", "<="),
        ("or before", "<="),
        ("and below", "<="),
        ("or below", "<="),
        ("and lower", "<="),
        ("or lower", "<="),
        ("and older", "<="),
        ("or older", "<="),
        ("and prior", "<="),
    ];
    for (phrase, operator) in bound_suffixes {
        if let Some(versions) = strip_suffix_ignore_case(range, phrase) {
            return bound_every_version(versions, operator, phrase);
        }
    }

    if let Some(version) = range.strip_suffix('+')
        && !version.contains([' ', ','])
    {
        return format!(">= {version}");
    }

    if let Some(requirement) = interval_to_requirement(range) {
        return requirement;
    }

    let range = replace_ignore_case(range, " up to and including ", " - ");
    let range = replace_ignore_case(&range, " through and including ", " - ");
    let range = replace_ignore_case(&range, " through ", " - ");
    let range = replace_ignore_case(&range, " until ", " - ");
    let range = replace_ignore_case(&range, " to ", " - ");
    replace_ignore_case(&range, " and ", ", ")
}

// ----- text helpers -----

// `str::strip_prefix` that ignores ASCII case and the whitespace after the prefix.
//   (`All versions prior to 1.7.14`, `all versions`)  ->  Some(`prior to 1.7.14`)
//   (`>= 1.0.0`, `all versions`)                      ->  None
fn strip_prefix_ignore_case<'a>(text: &'a str, prefix: &str) -> Option<&'a str> {
    let candidate = text.get(..prefix.len())?;
    if candidate.eq_ignore_ascii_case(prefix) {
        Some(text[prefix.len()..].trim_start())
    } else {
        None
    }
}

// `str::strip_suffix` that ignores ASCII case and the whitespace before the suffix.
//   (`1.20.1 and Earlier`, `and earlier`)  ->  Some(`1.20.1`)
//   (`1.20.1`, `and earlier`)              ->  None
fn strip_suffix_ignore_case<'a>(text: &'a str, suffix: &str) -> Option<&'a str> {
    let start = text.len().checked_sub(suffix.len())?;
    let candidate = text.get(start..)?;
    if candidate.eq_ignore_ascii_case(suffix) {
        Some(text[..start].trim_end())
    } else {
        None
    }
}

// `str::replace` that ignores ASCII case when searching.
//   (`1.2.0 Through 1.8.7`, ` through `, ` - `)  ->  `1.2.0 - 1.8.7`
//   (`a and b AND c`, ` and `, `, `)             ->  `a, b, c`
pub fn replace_ignore_case(text: &str, from: &str, to: &str) -> String {
    let lowercase_text = text.to_ascii_lowercase();
    let lowercase_from = from.to_ascii_lowercase();
    let mut result = String::new();
    let mut rest_starts_at = 0;
    for (index, _) in lowercase_text.match_indices(&lowercase_from) {
        result.push_str(&text[rest_starts_at..index]);
        result.push_str(to);
        rest_starts_at = index + from.len();
    }
    result.push_str(&text[rest_starts_at..]);
    result
}

// Replaces the characters people type instead of the semver ones. Markdown escapes, backticks and
// a trailing period are dropped.
//   `≤ 5.59.0`                 ->  `<= 5.59.0`
//   `=< 0.17.1`                ->  `<= 0.17.1`
//   `=>9.0.0, <= 9.0.2`        ->  `>=9.0.0, <= 9.0.2`
//   `>= 7.0.0-beta < =7.2.3`   ->  `>= 7.0.0-beta <=7.2.3`
//   `== 2.8.0`                 ->  `= 2.8.0`
//   `>= 6.19.4 && < 6.21.1`    ->  `>= 6.19.4 , < 6.21.1`
//   `<=7.1.0; <=6.9.11`        ->  `<=7.1.0, <=6.9.11`
//   `v0.10.0–v1.3.1`           ->  `v0.10.0 - v1.3.1`
//   `\>= 1.11.0, \<= 1.16.2`   ->  `>= 1.11.0, <= 1.16.2`
//   `<= v1.18.2` in backticks  ->  `<= v1.18.2`
//   `from 3.5.0 to 4.3.10.`    ->  `from 3.5.0 to 4.3.10`
pub fn replace_unusual_characters(range: &str) -> String {
    range
        .trim()
        .trim_end_matches('.')
        .replace('≤', "<=")
        .replace('≥', ">=")
        .replace("=<", "<=")
        .replace("=>", ">=")
        .replace("< =", "<=")
        .replace("> =", ">=")
        .replace("==", "=")
        .replace("&&", ",")
        .replace([';', '，'], ",")
        .replace(['–', '—'], " - ")
        .replace(['\\', '`'], "")
}